Dynamic Offloading for Multiuser Muti-CAP MEC Networks: A Deep Reinforcement Learning Approach

In this paper, we study a multiuser mobile edge computing (MEC) network, where tasks from users can be partially offloaded to multiple computational access points (CAPs). We consider practical cases task characteristics and capability at the CAPs may time-varying, thus, creating dynamic offloading problem. To deal with problem, first formulate it as Markov decision process (MDP), then introduce state action spaces. further design novel strategy based on deep Q network (DQN), dynamically fine-tune proportion in order ensure system performance measured by latency energy consumption. Simulation results are finally presented verify advantages of proposed DQN-based over conventional ones.